DATA SOCIETY®

Interactive visualization with R - Part 1

One should look for what is and not what he thinks should be.
-Albert Einstein.

Rising Data Science Interest

  • With the rise of COVID-19, the need for fast visualizations to put data into perspective has also risen
  • Have a look at this Visualization article while we wait to start class to get a peek at what's ahead.

centered

Recap

  • In the last module, we:

    • Transformed data using tidyverse
    • Plotted the transformed and tidy data to produce complex visualizations

  • In this module, we will create plots using the highcharter package

Module completion checklist

Objective Complete
Introduce using the highcharter package to build interactive visualizations
Use highcharter with tidy data to create a scatterplot
Visualize a correlation plot with hchart
Build a column plot with hchart
Create a boxplot with hchart
Save interactive plots with htmlwidgets

Directory settings

  • In order to maximize the efficiency of your workflow, you may want to encode your directory structure into variables
  • Let the main_dir be the variable corresponding to your skillsoft folder
# Set `main_dir` to the location of your `skillsoft` folder (for Mac/Linux).
main_dir = "~/Desktop/skillsoft"
# Set `main_dir` to the location of your `skillsoft` folder (for Windows).
main_dir = "C:/Users/[username]/Desktop/skillsoft"
# Make `data_dir` from the `main_dir` and remainder of the path to data directory.
data_dir = paste0(main_dir, "/data")
# Make `plots_dir` from the `main_dir` and remainder of the path to plots directory.
plot_dir = paste0(main_dir, "/plots")
# Set directory to data_dir.
setwd(data_dir)

Interactive visualizations with highcharter

  • Similar to ggplot2 and other plotting libraries (including base R itself), highcharter allows us to build complex, customized, and meaningful visualizations with the help of layers of graphical elements
  • Highcharter is an R wrapper that allows R users to tap into one of the most comprehensive data visualization JavaScript-based libraries: Highcharts
  • Highcharts is free for individual research and non-profit purposes, but there are cases and restrictions to its use and you may need to obtain a license if you decide to integrate it into software or company-wide products

centered-border

Interactive visualizations: highcharter

Let's install the package and check its documentation

# Install `highcharter` package.
install.packages("highcharter")

# Load the library.
library(highcharter)

# View documentation.
library(help = "highcharter")

centered-border

Highcharter main function: highchart

?highchart
  • Similar to ggplot2, in order to create a plot, we need to call the main plotting function ggplot()
  • In highcharter it is highchart()
  • The function doesn't need any required arguments, and all of the graphic parameters and plotting options can be specified within layers themselves

centered-border

Highcharter: `hchart` vs `highchart`

?hchart

hchart(Some_data,       #<- dataset to use
       "plot_type",     #<- plot type to use
       hcaes(x = variable1, #<- x-axis mapping 
             y = variable2, #<- y-axis mapping 
             group = variable3, #<- group by
             ...))

hchart is a shorthand version of the highchart function, which takes a few key arguments to plot:

  1. Data to use
  2. Type of plot to create (e.g., scatter, bar, column, line, etc.)
  3. hcaes (i.e., highcharts aesthetics) mapping of variables (works exactly the same way as with ggplot2!)

centered-border

Layers in Highcharts: series

  • Just like ggplot2, the highcharter library has its own vocabulary
  • Each new data / graphic layer in highcharter is called a series
  • Each series can be of different type. Here are some widely used ones:
Highcharter series type Plot type
scatter scatterplot
line line graph
boxplot boxplot
column bar plot
bar horizontal bar plot
histogram histogram
area density

Module completion checklist

Objective Complete
Introduce using the highcharter package to build interactive visualizations

✔
Use highcharter with tidy data to create a scatterplot
Visualize a correlation plot with hchart
Build a column plot with hchart
Create a boxplot with hchart
Save interactive plots with htmlwidgets

Set up: load & prepare data

  • Let's load the CMP_subset dataset from our data_dir into R's environment
# Set working directory to where we store data.
setwd(data_dir)

library(tidyverse)

# Read CSV file
CMP_subset = read.csv("CMP_subset.csv",
           header = TRUE)

Now, tidy the data as before and transform it from wide to long for easy visualization

# Prep data for univariate plots  
CMP_subset_long = CMP_subset %>%
  gather(key = "variable",
         value = "value") 

# Make names of processes and materials more user friendly and readable.
CMP_subset_long = CMP_subset_long %>%
     mutate(variable = 
            str_replace(variable,     
                          "Biological", "Bio ")) %>%  
     mutate(variable = 
            str_replace(variable, 
                          "Manufacturing", "Man. ")) %>%
     mutate(variable = 
            str_replace(variable, 
                          "0", " ")) %>%
    group_by(variable) %>%              #<- normalize
    mutate(norm_value =                 
           value/max(value, na.rm = TRUE))

Set up: prepare data

# Prep data for scatterplot
CMP_subset_long2 = CMP_subset %>%
  gather(BiologicalMaterial01:ManufacturingProcess03, 
         key = "variable",                            
         value = "value") %>%                         
  # All other transformations we've done before.
  mutate(variable = str_replace(variable, "Biological", "Bio ")) %>%
  mutate(variable = str_replace(variable, "Manufacturing", "Man. ")) %>%
  mutate(variable = str_replace(variable, "0", " ")) %>%
  group_by(variable) %>%
  mutate(norm_value = value/max(value, na.rm = TRUE))

head(CMP_subset_long2,3)

# A tibble: 3 x 4
# Groups:   variable [1]
  Yield variable       value norm_value
  <dbl> <chr>          <dbl>      <dbl>
1  38   Bio Material 1  6.25      0.709
2  42.4 Bio Material 1  8.01      0.909
3  42.0 Bio Material 1  8.01      0.909

hchart: scatterplot

To construct a scatterplot with highcharter, we use the hchart() function and pass the data, plot type ( scatter), and aesthetics to it as arguments

# Construct an interactive scatterplot.
scatter_interactive =              #<- name the plot   
  hchart(CMP_subset_long2,         #<- set data
         "scatter",                #<- plot type "scatter"
          hcaes(x = norm_value,    #<- set aesthetics to map x-axis
                y = Yield,         #<- set aesthetics to map y-axis
                group = variable)) #<- group by
  • In R, interactive charts appear in the Viewer pane, right next to the Help tab

centered-orange-border

hchart: scatterplot (cont'd)

scatter_interactive

Scatterplot: selecting categories

  • In highcharts every plotted category seen in the legend is called a series
  • Highcharts auto-colors series when it detects more than 1 category
  • We can interactively select and de-select which series to display by clicking on the series names in the legend

Scatterplot: `pipe` and customize

  • In highcharts every new option or layer can be added using the already familiar pipe operator (as opposed to the + operator in ggplot2)
  • The hc_chart function is responsible for global chart options like zoom, size, and theme
  • Let's set a zoom for our plot by passing the zoomType argument to hc_chart
  • xy zoom allows to zoom across both x and y axes
# Pipe chart options to original chart.
scatter_interactive = scatter_interactive %>%
  # Use chart options to specify zoom.
  hc_chart(zoomType = "xy") 

scatter_interactive

centered

Scatterplot: add title

A title can be added to highcharter plots using the hc_title() function

# Pipe chart options to original chart.
scatter_interactive = scatter_interactive %>%
 # Add title to the plot.
 hc_title(text = "CMP data: Yield vs. other variables")

scatter_interactive
As opposed to ggplot2 that uses a plus (+), highcharter uses a pipe (%>%) operator to feed objects and previous commands into the next ones. It works exactly the same way as we have already seen in tidyverse packages like dplyr and tidyr.

Module completion checklist

Objective Complete
Introduce using the highcharter package to build interactive visualizations

✔
Use highcharter with tidy data to create a scatterplot

✔
Visualize a correlation plot with hchart
Build a column plot with hchart
Create a boxplot with hchart
Save interactive plots with htmlwidgets

hchart: correlation matrix

  • hchart recognizes the type of data being given to it
  • If we pass a correlation matrix, it will create a correlation plot
  • No other arguments are necessary to create a basic plot!
# Compute a correlation matrix for the first 
# 4 variables in our data.
cor_matrix = cor(CMP_subset[, 1:4])

# Construct a correlation plot by 
# simply giving the plotting function
# a correlation matrix.
correlation_interactive = hchart(cor_matrix) %>%
 # Add title to the plot.
 hc_title(text = "CMP data: correlation")

hchart: correlation matrix (cont'd)

correlation_interactive

Knowledge check 1

centered

Add instructions for knowledge check here later

Exercise 1

centered

Module completion checklist

Objective Complete
Introduce using the highcharter package to build interactive visualizations

✔
Use highcharter with tidy data to create a scatterplot

✔
Visualize a correlation plot with hchart

✔
Build a column plot with hchart
Create a boxplot with hchart
Save interactive plots with htmlwidgets

hchart: column plot

Let's now create an interactive plot to visualize the summary of our data

  • To do that, we first need to get the summary statistics and save them as a dataframe
# Create data summary.
CMP_summary = summary(CMP_subset)

# Save it as a data frame.
CMP_summary = as.data.frame(CMP_summary)

# Inspect the data.
head(CMP_summary)

  Var1      Var2            Freq
1          Yield Min.   :35.25  
2          Yield 1st Qu.:38.75  
3          Yield Median :39.97  
4          Yield Mean   :40.18  
5          Yield 3rd Qu.:41.48  
6          Yield Max.   :46.34  

# Remove an empty variable.
CMP_summary$Var1 = NULL

# Rename remaining columns.
colnames(CMP_summary) = c("Variable", 
                          "Summary")
# Inspect updated data.
head(CMP_summary)

   Variable         Summary
1     Yield Min.   :35.25  
2     Yield 1st Qu.:38.75  
3     Yield Median :39.97  
4     Yield Mean   :40.18  
5     Yield 3rd Qu.:41.48  
6     Yield Max.   :46.34

Column plot: prepare data

  • Let's separate the summary values from the summary statistics
  • We can utilize the separate function in tidyr
# Separate `Summary` column into 2 columns.
CMP_summary = CMP_summary %>%              #<- set original data 
  separate(Summary,                        #<- separate `Summary` variable 
           into = c("Statistic", "Value"), #<- into 2 columns: `Statistic`, `Value`
           sep = ":",                      #<- set separating character
           convert = TRUE)                 #<- where applicable convert data (to numeric)

# Inspect the first few entries in data.
head(CMP_summary)

   Variable Statistic Value
1     Yield   Min.    35.25
2     Yield   1st Qu. 38.75
3     Yield   Median  39.97
4     Yield   Mean    40.18
5     Yield   3rd Qu. 41.48
6     Yield   Max.    46.34

# Inspect total number of rows in data including NAs.
nrow(CMP_summary)

[1] 49

Column plot: create plot

# Inspect `Value` column for `NAs`.
which(is.na(CMP_summary$Value) == TRUE)

[1]  7 14 21 28

# Subset only rows where `Value` is not NAs.
CMP_summary = subset(CMP_summary, !is.na(Value))

# Now the number of rows should be 4 less.
nrow(CMP_summary)

[1] 45

# Construct the summary chart.
CMP_summary_interactive = 
  hchart(CMP_summary,             #<- set data
         "column",                #<- set type (`column` in highcharts)
         hcaes(x = Statistic,     #<- arrange `Statistics` across x-axis
               y = Value,         #<- map `Value` of each `Statistic` to y-axis
               group = Variable)) #<- group columns by `Variable`

Column plot: display plot

CMP_summary_interactive

Column plot: customize tooltip

  • Since we are comparing variables' summary statistics, it would be convenient for the tooltip to contain information about the group rather than the individual columns within the group
  • We can control different tooltip options of the chart using the hc_tooltip option
  • The shared option is often used to share a tooltip between members of a group


# Adjust tooltip options by piping `hc_tooltip` to base plot.
CMP_summary_interactive = CMP_summary_interactive %>% 
  hc_tooltip(shared = TRUE)  %>%               #<- `shared` needs to be set to `TRUE`
  hc_title(text = "CMP data variable summary") #<- add title to your plot

Column plot: customize tooltip (cont'd)

CMP_summary_interactive

Module completion checklist

Objective Complete
Introduce using the highcharter package to build interactive visualizations

✔
Use highcharter with tidy data to create a scatterplot

✔
Visualize a correlation plot with hchart

✔
Build a column plot with hchart

✔
Create a boxplot with hchart
Save interactive plots with htmlwidgets

Highcharts boxplot: hcboxplot

hcboxplot allows us to create an interactive boxplot. It needs two arguments:

  • x requires the numeric data to be plotted along the x-axis (boxplot in highcharts is horizontal by default)
  • var requires categorical data to be plotted along the y-axis
?hcboxplot

hcboxplot(x = Numeric_data_vector,
          var = Categorical_data_vector,
          ...)

centered-border

Highcharts boxplot: hcboxplot (cont'd)

# Construct an interactive boxplot.
boxplot_interactive =  
  hcboxplot(x = CMP_subset_long$norm_value,
            var = CMP_subset_long$variable,
            name = "Normalized value") %>%
  hc_title(text = "CMP data variables")

boxplot_interactive

Highcharts plotOptions

  • To control individual layer/series options for various plot types, we use the hc_plotOptions function
  • It can be piped to the original chart to enhance our base plot
  • Each series type represented in the chart can be given a unique set of options
  • Every set of options is a list of micro-level adjustments

centered-border

Make sure to look up the proper function and option names in the highcharter documentation to make sure they match the original JS library!

hcboxplot: customize with `hc_plotOptions`

Now we will use the hc_plotOptions() function to customize color for our boxplot

# Enhance original boxplot with color options.
boxplot_interactive = boxplot_interactive %>% 
  hc_plotOptions(   #<- plot options
    boxplot = list(     #<- for boxplot 
      colorByPoint = TRUE)) #<- color each box

boxplot_interactive

Module completion checklist

Objective Complete
Introduce using the highcharter package to build interactive visualizations

✔
Use highcharter with tidy data to create a scatterplot

✔
Visualize a correlation plot with hchart

✔
Build a column plot with hchart

✔
Create a boxplot with hchart

✔
Save interactive plots with htmlwidgets

Save interactive plots: htmlwidgets

  • The htmlwidgets package lets us use JavaScript visualization libraries in R console
  • We can embed widgets in R markdown and Shiny web applications
  • Explore htmlwidgets here
# Install `htmlwidgets` package.
install.packages("htmlwidgets")

# Load the library.
library(htmlwidgets)

# View documentation.
library(help = "htmlwidgets")

centered-border

Save interactive plots: htmlwidgets (cont'd)

# Set working directory to where you save plots.
setwd(plot_dir)

# Save desired interactive plot to an HTML file.
saveWidget(scatter_interactive,        #<- plot object to save
           "interactive_scatterplot.html", #<- name of file to where the plot is to be saved
           selfcontained = TRUE)              #<- set `selfcontained` to TRUE, so that 
                                              #   all necessary files and scripts are embedded 
                                              #   into the HTML file itself

Knowledge check 2

centered

Add instructions for knowledge check here later

Exercise 2

centered

Module completion checklist

Objective Complete
Introduce using the highcharter package to build interactive visualizations

✔
Use highcharter with tidy data to create a scatterplot

✔
Visualize a correlation plot with hchart

✔
Build a column plot with hchart

✔
Create a boxplot with hchart

✔
Save interactive plots with htmlwidgets

✔

Summary

In this module, we:

  • Outlined the highcharter package for creating interactive visualizations
  • Built an interactive scatterplot, column plot, and boxplot using the hchart() function of highcharter
  • Customized plots using various highchart options and parameters
  • Learned to save interactive plots using htmlwidgets so that they can be embedded in R markdown and R shiny applications

In the next module, we will:

  • Introduce how to build an interactive map with a json file
  • Discuss best practices for highcharter

This completes our module

Congratulations!

centered